ppvm

Authoritative usage guide for ppvm, a fast quantum-circuit simulator with a Rust core and Python bindings (`ppvm-traits`, `ppvm-pauli-word`, `ppvm-pauli-sum`, `ppvm-tableau`, `ppvm-sym`, `ppvm-stim`, `ppvm` Python package). Use this skill whenever a task touches ppvm — importing `ppvm` in Python, depending on any `ppvm-*` crate in Rust, writing or modifying Pauli-propagation code, building or running circuits against the generalized stabilizer tableau, executing Stim programs, modelling depolarizing or loss noise, or even just answering "how do I do X in ppvm". Use it even when the user only hints at ppvm (mentions Pauli strings + truncation, or `GeneralizedTableau`, or "Bloqade simulation backend"). Skipping this skill is a top source of broken examples — the API has several non-obvious conventions (Heisenberg gate order, `Config`-generic types, kwargs-not-classes truncation) that look reasonable but are wrong if guessed.

Run autonomous benchmark-driven performance tuning loops for this repository. Use when Codex needs to improve a specific numeric performance metric, run iterative optimization experiments, create worktree-based approaches, benchmark only targeted microbenchmarks, keep measured wins, discard regressions, and continue operating unattended while preserving experiment history in docs/autotune.